American Finance Association
Interest Rates as Options
Author(s): Fischer Black
Source: The Journal of Finance, Vol. 50, No. 5 (Dec., 1995), pp. 1371-1376
Published by: Blackwell Publishing for the American Finance Association
Stable URL: http://www.jstor.org/stable/2329320
Accessed: 09/11/2010 02:44
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=black.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Blackwell Publishing and American Finance Association are collaborating with JSTOR to digitize, preserve
and extend access to The Journal of Finance.
http://www.jstor.org
THE JOURNALOF FINANCE . VOL.L, NO. 7 . DECEMBER1995
Interest Rates as Options
FISCHER BLACK*
ABSTRACT
Since people can hold currencyat a zero nominal interest rate, the nominal short rate
cannot be negative. The real interest rate can be and has been negative, since low risk
real investment opportunities like filling in the Mississippi delta do not guarantee
positive returns. The inflation rate can be and has been negative, most recently (in
the United States) during the Great Depression. The nominal short rate is the
"shadowreal interest rate" (as defined by the investment opportunity set) plus the
inflation rate, or zero, whichever is greater. Thus the nominal short rate is an option.
Longer term interest rates are always positive, since the future short rate may be
positive even when the current short rate is zero. We can easily build this option
element into our interest rate trees for backward induction or Monte Carlo simula-
tion: just create a distribution that allows negative nominal rates, and then replace
each negative rate with zero.
I. Interest Rate Processes
WHEN VALUING INTEREST RATE derivatives, we commonly model the nominal
interest rate process, though investors ultimately care more about the real
interest rate. When we model the nominal interest rate, we ignore the supply
of and demand for capital. We assume a simple random process, without
worrying about the forces that influence the interest rate.
We generally choose a normal process, a lognormal process, or a square root
process. With the normal process, the volatility of the change in the interest
rate does not depend on the rate, though it may depend on time. With the
lognormal, the volatility of the fractional change in the interest rate does not
depend on the rate. With the square root process, the ratio of the variance of
the change in the interest rate to the rate does not depend on the rate, so the
volatility of the change in the rate is proportional to the square root of the rate
at a given time. We can put mean reversion into any of these processes.
The normal process assumes that the interest rate volatility does not decline
as the rate approaches zero. This seems a little strange since volatility does
seem to decline as the rate declines, but it might be a good first approximation.
* Goldman, Sachs & Co. I am grateful to Chi-Fu Huang and Wei-TongShu for discussions of
these issues, and to Wei-Tong Shu for producing the graphs. [Note from the Managing Editor:
Fischer Black submitted this paper on May 1, 1995. His submission letter stated: "Iwould like to
publish this, though I may not be aroundto make any changes the referee may suggest. If I'm not,
and if it seems roughly acceptable, could you publish it as is with a note explaining the circum-
stances?"Fischer received a revise and resubmit letter on May 22 with a detailed referee's report.
He worked on the paper during the Summer and had started to think about how to address the
comments of the referee. He died on August 31 without completing the revision.]
1371
1372 The Journal of Finance
More seriously, though, the normal process allows the nominal interest rate to
become negative.
So long as people can hold currency, we know that the nominal short rate
cannot be negative. People would rather keep currency in their mattresses
than hold instruments bearing negative interest rates. In practice, the chance
of a negative interest rate in a process with mean reversion may be small, so
the values of derivatives may not depend much on cases where the rate is
negative. But the idea of a negative nominal rate is jarring.
The lognormal process implies that the nominal rate can never reach zero.
This seems a little strange, too, since the nominal short rate has sometimes
fallen to zero. In the United States, this happened most recently in the 1930s.
More generally, the lognormal implies that the volatility falls very rapidly as
the rate approaches zero. In most periods, while the volatility does seem to fall,
it does not seem to fall this rapidly.
The square root process is a compromise between the normal and the
lognormal, but is a little more complex than either of those processes. With the
square root process, when mean reversion is strong enough (or the drift in the
short rate is constant but large), the rate cannot reach zero. In other cases, the
rate can reach zero, and we must decide whether zero is a reflecting barrier or
an absorbing barrier. When zero is a reflecting barrier, the rate "bouncesoff'
zero, which seems strange. I can't think of any economic process that might
cause the short rate to bounce in this way. When zero is an absorbing barrier,
we must decide what determines the chance that the rate will become positive
again, and the model can become quite complex. Of the three choices, though,
the absorbing barrier seems most realistic.
Let us look more closely at why the short rate cannot be negative. It is
because currency is an option: when an instrument has a negative short rate,
we can choose currency instead. Thus, we can treat the short rate itself as an
option: we can choose a process that allows negative rates and can simply
replace all the negative rates with zeros. We still have a process with a single
number describing the state of the world: either the short rate (when it is
positive or zero) or what the short rate would be without the currency option
(when it is negative). We can call this number the "shadow short rate."
I find the behavior of a model like this realistic: the short rate can be stuck
at zero for a time, but its escape from zero is not random. We can follow the
"shadow short rate," and we know that whenever that becomes positive, the
short rate will be positive too.
We can use any of the three processes above as a base, but if we use a
lognormal or square root process, we will start it from a negative point. In
other words, we can use a "shifted"lognormal or square root process. Thus all
these processes can allow negative values for the shadow short rate, and
nonnegative values for the actual short rate once we add the option element.
Adding the nonnegative option to a process may kill any chance of having
analytic solutions for the prices of derivatives, but it hardly complicates nu-
merical solutions at all. If we use a tree, we simply start with an ordinary tree
Interest Rates as Options 1373
In
yield
o shadow yield>
0
ttn
0 10 20 30 40
maturity(years)
Figure 1. Zero-coupon yield.
for the shadow short rate, and replace any negative values with zeros before
valuing long-term securities or derivatives.
II. Yield Curves
Although the short rate can be zero while the shadow short rate is negative,
no longer term rate can be zero so long as there is a chance that the short rate
can become positive again. When we are valuing securities on a tree with finite
interval size, longer term rates may be zero on parts of the tree, but each rate
eventually becomes positive as we shrink the interval size. In the world, where
interval sizes are not relevant, all longer term rates should be positive in
equilibrium.
In effect, a forward rate is an option. The short rate at a future time is the
maximum of zero and the shadow rate, so it has an option-like value. The usual
factors affect the option value: increasing volatility or the current shadow rate
increases the forward rate. The effect of increasing maturity is ambiguous,
since the drift matters, and since "convexity" pushes the forward rate below
the expected short rate.
When the shadow rate is well above zero, especially when the interest rate
shows strong mean reversion, none of this matters much. But when the
shadow rate is near zero or negative, this causes the yield curve to slope
upward at first. As we increase maturity, the effective volatility increases, so
the forward rate increases (relative to what it would otherwise be).
To illustrate this, I have run a simple example with a normal distribution for
the shadow rate and no mean reversion. In Figure 1, we have the zero-coupon
yield curves for two cases: one where we allow the short rate to be negative,
and the other where we cut off the short rate distribution at zero. The shadow
short rate starts at 1 percent. It has zero drift in "risk-neutral space," and its
volatility is one percentage point. (Its variance rate is constant at .0001 per
year.)
Because of convexity, the yield curve tends to be concave downward. When
we put no floor on the short rate, the yield curve crosses zero and becomes
1374 The Journal of Finance
.............................
.................... . ....................
.................
shadow yield volatility~__--
6 ~~~~~~~yield
volatilit
0 10 20 30 40
maturity(years)
(short rate=1%;volatility=1%;mean reversion=0Q0)
Figure 2. Zero-coupon yield volatility.
negative at a maturity of 26 years. Putting a floor of zero on the short rate
causes the yield curve to rise at first. It reaches a maximum of 1.5 percent at
year 26 and then starts declining.
In Figure 2, we have the volatility curves for the same two cases. These are
yield volatilities in points, rather than fractional yield volatilities, which make
little sense when yields can be negative. Without the floor, yield volatilities
stay near one percentage point for all maturities. With the floor, the yield
volatility falls to about half a point at a maturity of 30 years.
Thus when the short rate is at or near zero, the floor has a big effect on both
the yield curve and the volatility curve.
III. Real Rates and Inflation
We define the real rate as the short-term nominal rate less the expected
inflation rate. We can define a "shadow real rate" as the shadow short-term
nominal rate less the expected inflation rate. WVhen the nominal rate is zero,
the shadow real rate has an economic meaning independent of any model. It is
what the real rate would be if we faced no floor on the nominal rate. For the
rest of this section, we will assume no floor on the nominal rate, so the real rate
and the shadow real rate are identical.
When the nominal rate has no floor, the real rate plays two roles: it helps
balance willingness to save and invest against opportunities to save and
invest, and it helps balance willingness to take risk across people. In an
unrestricted market, the cost of capital consists of the real interest rate and the
price of risk. (The price of risk is the extra expected return per unit of market
risk.) We cannot add these numbers, since they are not comparable. Thus the
cost of capital is always at least two numbers. In a restricted market, we need
even more numbers to describe the cost of capital. (A restricted market is one
where governments use quotas or taxes to regulate the flow of capital into or
out of specific countries or sectors.)
Interest Rates as Options 1375
Thus the real rate is at best only a part of the cost of capital. In an actual
economy, we have a term structure of real rates for each of several markets and
a term structure of prices of risk for each market.
Other things equal, when people become more willing to save and invest, the
real rate and the price of risk tend to fall. When saving and investment
opportunities become more attractive, the real rate and the price of risk tend
to rise. So long as people dislike risk, the price of risk must be positive. But
there is no similar reason why the real rate must be positive.
People will hold assets even when all expected returns are negative. They
may dissave and disinvest on balance, but they will continue to hold assets.
Tastes do not force the real rate to be positive.
Investment opportunities generally offer positive expected returns, which is
consistent with positive prices for risk, but need not imply a positive real
interest rate. If riskless real investment opportunities exist, they put a floor on
the real rate, but I don't know of any such opportunities. Martin Bailey once
claimed that filling in the Mississippi delta to create new farmland is a nearly
riskless investment with a positive return, but he failed to persuade me. That
investment looks risky to me, and I suspect its expected return is negative,
especially in today's environmentally sensitive climate. In fact, in most parts
of the delta, the investment would violate our "wetlands" laws.
People borrow and lend at the real rate to shift risk. Those who are more
tolerant of risk borrow, and those who are less tolerant lend. People who are
very averse to risk will lend even when the real rate is negative, so this does
not put a floor on the real rate either.
Moreover, in the years around 1980 the U.S. real rate was clearly and
consistently negative. Neither theory nor data suggest that the real rate must
be positive.
The inflation rate is a wholly different beast. The forces affecting the infla-
tion rate need not even overlap much with the forces affecting the aggregate
behavior of the real economy. Some see a significant tie between inflation and
phenomena like unemployment, but others see no tie at all.
In any case, the inflation rate need not be positive. Indeed, some have argued
that we should set the inflation rate to keep the nominal rate at zero, to reduce
the costs of economizing on currency balances. When the real rate is positive,
this means a negative inflation rate. Moreover, the expected inflation rate has
been negative at different times in U.S. history: most recently in the 1930s.
Since both the shadow real rate and the inflation rate can be negative, the
shadow nominal rate can be negative too. This means the floor at zero can be
binding.
IV. Equilibrium
What happens when the shadow nominal rate is negative? How does the
economy reach equilibrium?
When the government fixes a price, the exact method it uses affects the
resulting equilibrium. Buying unlimited amounts of corn at a fixed price gives
1376 The Journal of Finance
one equilibrium; maintaining the price by paying farmers not to plant corn
gives another.
When the government mandates that all trades must occur at a given price,
or above a certain minimum price, the economy must find a way to adapt to the
resulting excess supply. The minimum wage, for example, causes workers to
drop out of the "official"workforce. Many join the underground economy.
Meanwhile, employers ration jobs. They rearrange their production pro-
cesses to make less use of low skill labor. Speed in responding to a new job
opening becomes crucial among those with few skills. Employers may even use
the equivalent of lotteries to hire at these levels.
Similar issues arise when the shadow short rate is negative. By making
currency available, the government fixes the nominal short rate at zero. The
real rate is higher than the shadow real rate, so the government is "losing
money" on its short term liabilities. Seignorage has turned from a source of
profits to a source of losses.
At the artificially high real rate, the private sector wants to lend a lot but
borrowonly a little. It has an excess supply of lending. The government makes
up the difference by issuing currency, bills, notes, and bonds.
Investment opportunities that would be attractive at the shadow real rate
become unattractive at the actual real rate. The private sector disinvests,
perhaps excessively, unless the government steps in to subsidize these invest-
ments. Or the government may simply buy the businesses from the private
sector and operate them at a loss. The taxpayers make up the difference.
Since governments are usually less efficient than the private sector in
operating businesses, and since tax rates must be higher than normal in this
scenario, the economy may face substantial welfare losses. If governments fail
to see what's happening and don't subsidize or operate the businesses, the
economy may face massive unemployment and welfare losses.
Low-risk investments are displaced more than high-risk investments, since
the primary distortion is to the real rate rather than the price of risk. The
average risk of the private sector's investments goes up. This makes the
economy less stable.
The last time we faced this situation in the United States was in the 1930s.
The nominal short rate was near zero, and longer term rates were higher than
their shadow values because of the option providedby currency. (This is closely
related to what some macroeconomists call the "liquidity trap.") I claim this
contributed, perhaps a lot, to the severity of the Great Depression.
